The Phenotype and Neural Correlates of Language in Autism: an Integrative Review

Neuroscience and Biobehavioral Reviews 32 (2008) 1416–1425

Contents lists available at ScienceDirect

Neuroscience and Biobehavioral Reviews

journal homepage: www.elsevier.com/locate/neubiorev

Review The phenotype and neural correlates of language in autism: An integrative review

Wouter B. Groen a,b,c,*, Marcel P. Zwiers b,c, Rutger-Jan van der Gaag a,b, Jan K. Buitelaar a,b a Karakter, Child and Adolescent Psychiatry University Centre, Nijmegen, The Netherlands b Department of Psychiatry, Radboud University Nijmegen Medical Centre, Reinier Postlaan 10, 6500 Nijmegen, The Netherlands c FC Donders Centre for Cognitive Neuroimaging, Radboud University Nijmegen, Nijmegen, The Netherlands

ARTICLE INFO ABSTRACT

Keywords: Although impaired communication is one of the defining criteria in autism, linguistic functioning is Autism highly variable among people with this disorder. Accumulating evidence shows that language Language impairments in autism are more extensive than commonly assumed and described by formal diagnostic Connectivity criteria and are apparent at various levels. Phenotypically, most people with autism have semantic, Brain syntactic and pragmatic deficits, a smaller number are known to have phonological deficits. Phenotype Neural Neurophysiologically, abnormal processing of low-level linguistic information points to perceptual difficulties. Also, abnormal high-level linguistic processing of the frontal and temporal language association cortices indicates more self-reliant and less connected neural subsystems. Early sensory impairments and subsequent atypical neural connectivity are likely to play a part in abnormal language acquisition in autism. This paper aims to review the available data on the phenotype of language in autism as well as a number of structural, electrophysiological and functional brain-imaging studies to provide a more integrated view of the linguistic phenotype and its underlying neural deficits, and to provide new directions for research and therapeutic and experimental applications. ß 2008 Elsevier Ltd. All rights reserved.

Contents

1. Introduction ...... 1417 2. Methods...... 1417 3. The phenotype of language disorders in autism ...... 1418 3.1. Phonology ...... 1418 3.2. Semantics ...... 1418 3.3. Syntax...... 1418 3.4. Pragmatics and prosody...... 1419 4. The neural correlates of language disorders in autism ...... 1419 4.1. Structural abnormalities ...... 1419 4.2. Auditory perception and lower-level language paradigms...... 1420 4.3. Higher-level language paradigms ...... 1421 5. Discussion ...... 1421 5.1. The relationship between linguistic impairments and social and restricted/repetitive impairments in autism ...... 1422 5.2. Genetic, endophenotypical and behavioural correspondence between autism and language impairments ...... 1422 5.3. Implications for therapies and directions for future research ...... 1423 References ...... 1423

* Corresponding author at: Department of Psychiatry, Radboud University Nijmegen Medical Centre, Reinier Postlaan 10, 6500 Nijmegen, The Netherlands. Tel.: +31 24 3512222; fax: +31 24 3512211. E-mail address: [email protected] (W.B. Groen). Abbreviations: AS, Asperger syndrome; ASD, autism spectrum disorder; DTI, diffusion weighted imaging; ERP, evoked response potentials; fMRI, functional magnetic resonance imaging; HFA, high-functioning autism; MEG, magnetoencephalography; MMN, mismatch negativity; PDD-NOS, pervasive developmental disorder not otherwise speciﬁed; SLI, speciﬁc language impairment; ToM, theory of mind; WCC, weak central coherence.

1. Introduction ment in ToM in autism therefore causes an inability to comprehend the meaning of words. Furthermore, acquisition of language may Autism is a neurodevelopmental disorder associated with be mediated by shared or joined attention, which, in case of an lifelong handicaps in social adaptation. The disorder is defined by a impaired ToM, would be impaired as well (Kuhl et al., 2003). description of symptoms in three domains: social interaction, Semantic ability and false belief have indeed been found to verbal and non-verbal communication and stereotyped behaviour correlate in children with autism (Tager-Flusberg, 2000). (American Psychiatric Association, 1994). Autism is currently The psychological framework provided by these top-down considered to be one of the autism spectrum disorders (ASD), theories assumes that an impaired high-level cognitive function is which also include the Asperger syndrome (AS) and pervasive causing the impairments in autism. This assumption has been developmental disorder not otherwise specified (PDD-NOS). The criticised for several reasons. Firstly, converging evidence suggests term spectrum refers both to the heterogeneity of the ASD disorders that abnormalities of the processing of low-level sensory and the wide variety in functional ability. Although not universally information may lead to impairments in higher-order cognitive agreed upon, the definition of AS provided in DSM-IV entails functions, rather than the other way around (Happe and Frith, impaired social interaction and stereotyped behaviour in the 2006; Bertone et al., 2005). That is to say, altered low-level absence of language and cognitive delay (Klin et al., 2005). perceptual processing in autism should not be considered a by- Individuals with PDD-NOS fail to meet the full criteria for autism. product of weak central coherence. Quite on the contrary, This is the case when the number of criteria met is sub-threshold, perceptual abnormalities give rise to weak central coherence. or the onset occurs over the age of 3, or atypical symptoms are Secondly, these neuropsychological top-down theories are present, or a combination of the above (Buitelaar and van der Gaag, descriptive rather than explanatory, and finding the neural 1998; Buitelaar et al., 1999). correlates of these theories has been proven difficult since their Linguists commonly describe language and language disorders predictions of cortical functioning are too general to be falsifiable. in terms of phonology, semantics, syntax and pragmatics, and since Nevertheless, there have been converging efforts from different the early seventies, researchers have been analysing linguistic disciplines to document the neural correlates underlying the deficits in autism using these categories (Cromer, 1981). Whereas symptoms of ASD (Volkmar et al., 2004). Although most of the phonology deals with the perception and production of sound research has focused on impairments in social cognition (mostly units whose concatenation produces words, semantics deals with using visual stimuli such as faces), language impairments in autism the meaning of lexical items, syntax with the structure of words in are increasingly recognised. More and more findings on the clinical sentences, and pragmatics with the conventions and rules phenotype and the neural substrates of language and commu- governing the use of language for communication (Boucher, nication in autism are being added to the literature, but the broad 2003). From a neuroscientific perspective, however, there is no field of autism research and the many different analytical clear-cut relation between linguistic categories and cortical approaches make it difficult to oversee the current literature. This function. Recent functional imaging studies have furthered the paper therefore aims to review recent evidence from structural, idea that the language system is organised in a large number of electrophysiological and functional studies on the neural corre- small but tightly clustered modules in both the left and right lates of linguistic abnormalities in autism. Findings on the hemisphere with unique contributions to language processing. phenotype of language impairments in autism will also be There is also increasing evidence that cortical language regions are addressed here. We will argue that the linguistic features in not specific to language, but involve more reductionist processes autism cover a wider range of impairments than described in the that give rise to language as well as non-linguistic functions DSM-IV criteria for autistic disorder and are more linked to the (Bookheimer, 2002). neural architecture in autism than earlier behavioural studies have Given the absence of a clear-cut relation between the linguistic suggested. More specifically, language ability varies greatly among categories and cortical function, it is unfeasible to deduce people with autism. Although most individuals with autism have neurobiological deficits in autism from the high-order language semantic, syntactic and pragmatic language deficits, there is also a deficits characterising the disorder. Instead, several psychological number that have phonological difficulties. Functional brain- theories, such as the weak central coherence (WCC) theory (Frith, imaging data show aberrant neural activation in semantic, 1996) and the impaired theory of mind (ToM) (Baron-Cohen et al., syntactic and pragmatic tasks of higher-order language functions, 1985), have attempted to explain the high-order language deficits as well as in low-level sensory processes. Furthermore, we will in autism. The weak central coherence theory predicts that, since argue that the abnormalities of low-level sensory processing of people with autism are biased towards local versus global linguistic stimuli can be interpreted in the light of connectivity processing, their ability to integrate contextual information into models in autism. Finally, we will discuss the relationship between a composite whole is diminished. The high-order core deficit in language impairments and the other functional impairments in central processing supposedly results in altered low-level proces- autism (social interaction and stereotyped and rigid behaviour sing. Several studies have indeed demonstrated a reduced ability to patterns), as well as the relationship between autism and specific infer word-meaning from sentence context (Happe, 1997)orto language impairment (SLI), to provide a more integrated view of infer global meaning from sentences (Jolliffe and Baron-Cohen, the linguistic phenotype and its underlying neural deficits. An 2000), yielding empirical evidence for the WCC account for at least integrated review may be useful for both clinicians and research- the semantic and pragmatic language deficits in autism. However, ers, as it will allow further congruity between the observed WCC would also predict a superior performance on single word language deficits and their putative causes and could lead to tasks, as is the case in hyperlexia. Yet, hyperlexia is only rarely seen important therapeutic and experimental applications in the future. in autism. The majority of people with autism have difficulties with the meaning of isolated words as well as whole sentences. 2. Methods ToM refers to the specific cognitive ability to infer other people’s mental states and to understand that others have beliefs, We conducted an extensive internet search of the English desires and intentions that are different from our own. It has been literature published on the MEDLINE and PsycInfo databases in the argued that early stages of ToM are necessary for the ability to use past two decades, using the keywords (autism, autistic disorder, symbols such as words (Tager-Flusberg, 2000), and that impair- and Asperger syndrome) AND (language, language disorders, 1418 W.B. Groen et al. / Neuroscience and Biobehavioral Reviews 32 (2008) 1416–1425 language development disorders, language profile, communica- functioning autism (LFA) and least severe in those with HFA and AS tion, speech, voice, prosody, intonation, and auditory) AND (Boucher, 2003). Delay in speech acquisition is one of the primary (comprehension, brain, brain mapping, cerebral cortex, cortical, diagnostic characteristics and the degree of language impairment neural, hemisphere, functional, imaging, neuroimaging, magnetic is a key prognostic factor (Lord and Paul, 1997; Venter et al., 1992). resonance imaging (MRI), magnetoencephalography (MEG), elec- Yet, the course and development of semantic difficulties is an troencephalography (EEG), positron-emission tomography (PET)). under-researched area in autism. One comprehensive study As it is impossible to discuss all the publications here, we confined revealed marked difficulties in lexical comprehension and ourselves to reviewing the most recent representative papers in expressive vocabulary in the majority of children with autism each area that were published in peer-reviewed journals. (Kjelgaard and Tager-Flusberg, 2001). Furthermore, a strong correlation was found between full-scale IQ and performance on 3. The phenotype of language disorders in autism tests for comprehension (Peabody Picture Vocabulary Test) and expression (Expressive Vocabulary Test) (Dunn and Dunn, 1997; Impaired language function is frequently observed in people Williams, 1997). This correlation suggests that semantic compre- with autism, often in combination with mental retardation. hension and expression are unimpaired in the most able people Language dysfunction in autism is, however, much more variable with autism, and especially in those with Asperger syndrome. than the universal deficits in communication (Kjelgaard and Tager- Subtle semantic impairments are however also present in HFA and Flusberg, 2001). At one end of the autism spectrum there are AS. Howlin et al. found a poor performance on tests for productive children whose verbal abilities are within the normal range of (British Picture Vocabulary Scale) and receptive (Expressive One functioning, and at the other there are some who never start to Word Picture Vocabulary Test) semantic abilities for both AS and speak (Lord and Paul, 1997). In those with sufficient language and HFA (Gardner, 1982; Howlin, 2003; Dunn et al., 1997). While there cognitive abilities, i.e. people with high-functioning autism (HFA) was no significant difference for language comprehension between and Asperger syndrome, social communicative abilities remain HFA and AS, ratings for language expression revealed a small but impaired. Language tends to be used one-sidedly, non-reciprocally significant difference favouring the AS group. Thus, in accordance and instrumentally rather than for social purposes (Fine et al., with clinical experience and anecdotal evidence, language abilities 1994). In their original descriptions of ASD, Kanner and Asperger in people with HFA and AS differ mainly with respect to expressive both gave account of the typical language anomalies encountered abilities. in autism such as echolalia, pronoun reversal, utterances not It has been suggested that the semantic difficulties in autism related to the conversational context, and a lack of drive to engage are a consequence of deficits in advanced conceptualisation, since in communication (Asperger, 1991; Kanner, 1943). Researchers simple concepts referred to by lexical items are not affected in have subsequently elaborated the phenotype of language impair- autism (Tager Flusberg, 1981) and comprehension of terms ment in autism, distinguishing phonological, syntactic, semantic referring to emotional states (Tager Flusberg and Sullivan, 1995; and pragmatic deficits. Hobson and Lee, 1989) or abstract terms (Frith and Snowling, 1983) is more affected than comprehension of concrete words. 3.1. Phonology Other studies, however, do not support this finding. Interference during a reading task was not different for concrete and abstract It has been a widely held belief that the development of words (Eskes et al., 1990), and abstract terms that exist through phonology progresses at a slower rate, but is not impaired in human agency such as ‘war’ and ‘peace’ were not found to be used autism since phonologically correct echolalia is commonly found anomalously in autism (Perkins et al., 2006). Thus, the exact nature in low-functioning autism, suggesting that phonological percep- of the semantic deficits in autism remains to be established. tion and production is intact even in severely affected individuals (Tager Flusberg, 1996). However, there is converging evidence for 3.3. Syntax an articulation deficit in a subgroup of autistic children. A delayed developmental trajectory of phonology has been reported (Barto- As with the phonological properties of language in autism, lucci et al., 1976), as well as a greater number of articulation syntactic impairments have not been well researched. Recent distortion errors in people with HFA and the Asperger group than findings, however, do provide evidence for distinct and specific in typically developing speakers (Shriberg et al., 2001). Recently, syntactic deficits in autistic children who acquire spoken language, Kjelgaard and colleagues found phonological processing deficits and more noteworthy, they also show that, in those who do not (as measure by repeating non-sense words) in autistic children, acquire speech, the ability to acquire the grammar (and although only in those with concurrent impaired vocabulary and vocabulary) of signed language is impaired as well (Boucher, higher-order semantic and syntax deficits (Kjelgaard and Tager- 2003). The distinct syntactic deficits in autism entail reduced Flusberg, 2001). This language profile was also found by another expressive and receptive syntactic abilities, which have been found group in half of the children with poor communicative abilities using the Clinical Evaluation of Language Fundamentals (CELF) (Rapin and Dunn, 2003). It thus seems that, in a subgroup of (Kjelgaard and Tager-Flusberg, 2001) and mean utterance length in autistic children, phonology is impaired along with language free play sessions (Eigsti et al., 2007). Contrary to the consistent comprehension and production. This clustering of symptoms may findings on the length of syntactically complex utterances, the suggest that common causes underlie phonology and syntactic findings on grammatical morphemes such as verb tense markers abilities in autism and also in other disorders such as specific and articles are inconsistent. Whereas earlier studies found more language impairment. This will be dealt with in more detail in grammatical errors (Bartolucci and Albers, 1974; Bartolucci et al., Section 5. 1980), a more recent well-matched study found no such errors in spontaneous autistic speech (Eigsti et al., 2007). Yet, Roberts et al. 3.2. Semantics did find high rates of omissions of tense marking in a subgroup of children who were language impaired (Roberts et al., 2004). The Difficulties in both understanding and expressing lexicon are inconsistency of findings on grammatical morphemes probably the most widely recognised linguistic impairments in autism. reflects the fact that subgroups are affected rather than all autistic Semantic impairments are most severe in people with low- people, so that sufficiently large subject groups are needed. W.B. Groen et al. / Neuroscience and Biobehavioral Reviews 32 (2008) 1416–1425 1419

It has been a long-held belief that people with autism are better A pragmatic ability closely related to language comprehension at language production than at language comprehension. The main is the ability to perceive and use intonation, rhythm, tone of voice reason for this assumption was a series of studies conducted in the and stress, referred to as prosody. The use of aberrant prosody is seventies that compared autistic children with children with mentioned in the DSM-IV description of autism, and the perception severe receptive language disorder. These studies revealed of prosody in autism has been investigated by several authors. significant differences in comprehension of vocabulary and the Rutherford et al., for example, compared the ability of adults with production of syntactically complex utterances, favouring the ASD to attribute emotions to sentences spoken with an emotional children with receptive language disorder (Bartak et al., 1975; Cox tone of voice, and found that individuals with the HFA or Asperger et al., 1975; Cantwell et al., 1978). It was therefore argued that syndrome have difficulties extracting mental state information syntactic comprehension was more affected than production. from vocalisations (Rutherford et al., 2002). Researchers found However, in contrast with the above finding for syntax, a more deficits in the perception and production of stress, intonation and recent study did not replicate a difference between semantic phrasing (Paul et al., 2005), as well as a preference for non-verbal comprehension and expression (Kjelgaard and Tager-Flusberg, sounds and an indifference to the mother’s voice in children with 2001). A language profile in autism with better production than autism (Dawson et al., 1998; Klin, 1991). The latter finding seems comprehension could thus not be replicated. Nevertheless, to indicate that deficits in the perception of prosody do not reflect evidence was found for two distinct language profiles in autism. an innate inability, but arise as a consequence of non-social The language profile in one subgroup entailed worse performance orientation. However, the direction of causality may also be the on tests of grammatical ability than vocabulary (Kjelgaard and other way around, or even be bidirectional. The contribution of Tager-Flusberg, 2001), possibly clustering with phonological low-level deficits to high-level deficits will be dealt with in greater impairments (Rapin and Dunn, 2003) (see also Section 3.1). The detail in Section 5. language profile in the other subgroup entailed impaired In summary, linguistic and non-linguistic pragmatic deficits semantics and pragmatics (Rapin and Dunn, 2003). These subtypes and prosodic deficits are part of the most commonly affected suggest that syntactic and phonological deficits have a common domains of functioning measured across the spectrum of autism cause, as have semantic and pragmatic deficits in autism. However, disorders. It appears that linguistic abilities do indeed facilitate the distinction of subtypes was based on a number of studies with pragmatic abilities, although they are not necessary per se for an autistic sample that was pre-selected for the presence of pragmatic competence. difficulties in language comprehension, so the extent to which these findings can be generalised to the broad autism spectrum 4. The neural correlates of language disorders in autism remains to be established. In summary, syntactic abilities in autism are characterised by Although autism is a heterogeneous disorder that includes sparse expressive language with immature syntax in a majority of many contradictory neurophysiological findings, several neural children. There is some evidence for a clustering of syntactic/ correlates underlying the linguistic deficits of autism have been phonological deficits in a subgroup of children and a clustering of reproduced in different investigative modalities. Some researchers semantic/pragmatic deficits in a subgroup, but this will require focused on structural abnormalities (Herbert et al., 2002), while confirmation in an unselected autistic population. others addressed the perception of simple linguistic stimuli (Boddaert et al., 2003; Ceponiene et al., 2003) and higher-level 3.4. Pragmatics and prosody linguistic functions (Harris et al., 2006; Kana et al., 2006). The results indicate that individuals with autism activate alternative Pragmatics entails both the linguistic and non-linguistic items and possibly less flexible networks during phonetic, semantic, that are covered by the defining criteria of autism. Linguistic syntactic and pragmatic language processing. Abnormalities at an pragmatics implies difficulties in the ability to disambiguate early level of information processing contribute to the hypothesis meaning, the ability to structure coherent discourse and to of a bottom-up aetiology in which, for example, an alteration of understand irony and implied meaning. The ability to understand auditory cortical processing leads to abnormal language develop- other people’s intentions, social rules of conduct and non-verbal ment in early childhood. communication gestures are regarded as non-linguistic pragmatics. Deficits in pragmatic functioning are evident at all 4.1. Structural abnormalities developmental stages, even in highly verbal adults with autism (Lord and Paul, 1997; Tantam et al., 1993; Happe, 1993; Martin and After a long period of inconsistent findings in structural McDonald, 2004; Baron-Cohen, 1997). neuroimaging studies, data of recent studies converge to elucidate It should be noted that most research in autism on non-linguistic the underlying abnormalities in autism (see Palmen and van pragmatic abilities, such as the ability to understand other people Engeland, 2004 for a review). Using MRI, an abnormal develop- intentions, as measured by ToM tests, involve language. Data on mental trajectory of the cerebral cortex in autism was found, in whether the deficit in ToM extends to people with autism with little which autistic children had normal head circumference at birth, or no language is sparse, making it difficult to disentangle the and increased grey and white matter volume at 2–3 years of age, relative contributions of linguistic and non-linguistic deficits in the which normalised again in later years (Courchesne et al., 2001). socio-communicative deficits in autism. Moreover, it has been Particularly the frontal lobes were abnormally enlarged (Carper argued that the syntactic ability to build subordinate clauses allow et al., 2002). The abnormal developmental trajectory was children to reason about mental states that are at odds with reality confirmed in a meta-analysis of head circumferences, post- (e.g. ‘‘Jane thinks the cookies are in the cabinet’’) (De Villiers and De mortem findings and MRI measurements (Redcay and Courchesne, Villiers, 1995). Syntax mastery has indeed been found to correlate 2005), which led the authors to argue that the early overgrowth with performance on ToM tasks. However, using a non-verbal ToM interferes with the normal developmental trajectory of cortical test, Colle et al. found that children with autism selectively failed connectivity. Cellular abnormalities such as neuroinflammatory false-belief tasks, whereas children with SLI or typically developing processes (Vargas et al., 2005) and supernumerary neurons (Smitz, children did not (Colle et al., 2007), indicating dissociation between 2005) that are predominantly present in the frontal, temporal and verbal and pragmatic abilities. cerebellar structures (Courchesne et al., 2005) may lead to an 1420 W.B. Groen et al. / Neuroscience and Biobehavioral Reviews 32 (2008) 1416–1425 impairment in the information transfer between the frontal cortex delayed magnetic mismatch field for vowels (but not for tones) by and other systems (Courchesne and Pierce, 2005). The hypothesis using magnetoencephalography (Kasai et al., 2005). Thus, early of aberrant collaboration between cortical areas or abnormal perceptual processing in autism appeared to be deficient and cortical connectivity in autism is a scientific area that is currently speech-specific. Others studies have confirmed the early percep- receiving much attention. In Section 5, we will go deeper into its tual processing deficits, but the speech specificity was not implications for language in autism. replicated: they reported smaller N1c waves in children with More importantly, the areas associated with cellular abnorm- autism in an event-related paradigm with simple tones (Bruneau alities in autism (the frontal, temporal and cerebellar structures) et al., 1999) and delayed MMN for both speech and non-speech all subserve language functions, although the cerebellum is only sounds (Oram Cardy et al., 2005). Whether or not cortical sound considered to play a facilitating role (Allen et al., 2004). processing impairments are speech-specific, sound processing Consequently, morphometric findings on the frontal and temporal impairments may be fundamentally associated with language language areas have been associated with language impairments impairments. in autism. Asymmetry reversal of the frontal language-related In an fMRI study that contrasted voices with environmental cortex was found (DeFosse et al., 2004; Herbert et al., 2002; Abell sounds, no differential activation between the two conditions was et al., 1999), as well as anterior and superior shifting of the left found in the autism group, suggesting that speech was not processed inferior frontal sulcus and superior temporal sulcus bilaterally by a speech-specific cortical region (Gervais et al., 2004). The control (Levitt et al., 2003) and decreases of grey matter concentration group, on the other hand, showed greater activation for voices along bilaterally in the superior temporal sulcus (Boddaert et al., 2004b). the upper bank of the superior temporal sulcus bilaterally. The Although cognitive functions are difficult to relate to morpho- abnormal pattern of cortical activation in the autism group might metric findings in a one-to-one manner, the main conclusion of reflect a bottom-up sensory impairment, or alternatively, it may be these studies is that the cortical development of language-related caused by an attention bias towards non-vocal sounds, which in turn areas follows a different trajectory in autism, possibly in the may lead to the development of linguistic deficits in autistic context of a reduced left-lateralised hemispheric dominance. children. This problem was tackled by a positron emission Whether these findings of aberrant development and collaboration tomography (PET) study that investigated the auditory cortical between cortical areas are specific for autism is, however, not clear. processing of pre-linguistic speech-like sounds that have an acoustic As a consequence, there is a need to directly compare the structure similar to speech (Boddaert et al., 2003; Boddaert et al., neurobiological development and functioning in autism with other 2004a). The study reported a reversed hemispheric dominance in disorders of language such as SLI and dyslexia. the autism group and, compared with controls, less left-temporal activation and greater right middle frontal gyrus activation. The 4.2. Auditory perception and lower-level language paradigms elegance of the study design was that top-down processing of language was less likely because the sounds were not perceived as Several authors have focused on clarifying two main questions language but rather as strange electronic tones. The acoustic in the field of language in autism: whether or not early sensory structure was made to resemble speech sounds, such that any processes are deficient in autism and whether or not these early cortical processing differences reflected bottom-up linguistic sensory processes are speech-specific. These questions are highly abnormalities. The authors therefore speculated that the observed relevant to better understand the nature of language deficits in abnormal auditory processing led to an abnormal early stage of autism, since they allow inferences on the contribution of top- language development rather than that it reflects the consequences down versus bottom up-processes in the functional impairments of abnormal language development. in autism. Basically, bottom-up theories assume that early sensory The above-mentioned studies converge to the idea that deficits give rise to deficient high-level operations such as impairments of information processing may lead to an abnormal language and social behaviour (cf. the inability to speak in deaf developmental trajectory of language in autism. Studies of the people). Alternatively, top-down theories assume that the ability maturational path of cortical language processing do indeed to perceive stimuli is not impaired. Yet, deficits in high-level indicate an abnormal developmental trajectory in the form of processes such as social behaviour could give rise to undirected reversed hemispheric dominance. Dawson et al. reported a strong attention, or attention directed to non-socially relevant clues, so relationship between language ability and hemispheric asymme- that word-object associations do not develop efficiently in autistic try in ERP, on the one hand, and vowel stimuli in autism, on the children. In healthy children, joint attention has indeed been other (Dawson et al., 1989). Furthermore, in a MEG study using shown to be a prerequisite for acquiring language (Kuhl et al., simple vowel stimuli, Flagg et al. found rightward instead of typical 2003), suggesting a role for top-down influences in autism. Yet, it is leftward hemispheric lateralisation (Flagg et al., 2005). Reduced still possible that sensory deficits give rise to deficits in joint leftward hemispheric activation was also found in non-linguistic attention. To shed light on this chicken-and-egg problem, much neuroimaging studies (Gendry-Meresse et al., 2005; Chiron et al., data has been collected using neurophysiological experiments that 1995). Yet, the association of cortical processing impairments and measure neural responses to stimuli at a millisecond scale. impaired language in autism does not per se imply causality. That Cˇeponiene` et al. used an ERP oddball paradigm to examine the is, cortical sound processing impairments may cause language sensory and early attentional processing of speech and non-speech deficits, but language deficits may also cause cortical sound sounds in children with HFA (Ceponiene et al., 2003). They found processing impairments. Alternatively, another factor such as a no differences in mismatch negativity (MMN, an index of genetically programmed maldevelopment of neuroarchitecture in automatic sound change detection, situated in time after the autism may give rise to both cortical sound processing impair- N1c wave and before the P3a wave), which suggests intact early ments and language deficits. perceptual abilities in autism. However, involuntary orienting In short, impaired cortical sound processing is reproduced (P3a) was different for speech sounds but not for non-speech reliably in autism. Whether or not the auditory processing sounds in the HFA group. This finding supports the hypothesis of a abnormalities are speech-specific is still unclear, but early sensory speech-specific post-sensory auditory impairment, suggesting that impairments and their interplay with the maturational path of the people with autism may perceive but not attend to linguistic cerebral cortex are likely to play a part in the abnormal language stimuli. Contrary to Cˇeponiene`’s findings, Kasai et al. did find a acquisition in autism. W.B. Groen et al. / Neuroscience and Biobehavioral Reviews 32 (2008) 1416–1425 1421

4.3. Higher-level language paradigms Thus, the neuroimaging data give rise to four main conclusions. Firstly, the data indicate that people with autism tend to rely more Given the neuroimaging data found so far, it is unlikely that a on Wernicke’s area (temporal region) and less on Broca’s area deficit in a single cortical area can account for the phenotypic (frontal region) for processing sentences and single words. Since language deficits in autism, as is the case in post-stroke aphasia. In Broca’s area subserves integration processes (Hagoort et al., 2004) the autistic brain, cortical regions seem to collaborate in a different and Wernicke’s area is more associated with semantic retrieval, the fashion, and cortical areas that show hypoactivation or hyper- neuroimaging data present an intriguing parallel between the activation for one task may show normal activation for another WCC’s piecemeal processing style and the greater reliance on task. Wernicke’s area neurally. It might thus well be that early sensory The first functional imaging study of higher-level language processes that highlight individual components in a composite perception and generation in autism was a small exploratory PET whole give rise to the greater Wernicke’s area activation rather study conducted in 1999 (Muller et al., 1999). In five autistic than that Wernicke’s area itself is malfunctioning. participants, sentence perception was associated with the reversal Secondly, the neuroimaging data suggest that, during high-level of normal left-hemisphere dominance, while sentence generation processing in autism, cortical areas are used in a different manner showed normal left inferior frontal activation. In a re-analysis of and in reaction to other stimuli rather than that certain cortical the data on the same subjects, which focused on three regions of areas are malfunctioning per se. An elegant example of this are two interest, participants with autism had a reduced Broca’s area studies by the same group that assessed the neural basis of irony (Broca’s area is situated in the opercular and triangular sections of comprehension in autism. One study found hypoactivation of the inferior frontal gyrus) activation during language perception prefrontal and temporal regions during judgment of scenarios that and production (Muller et al., 1998). A more recent fMRI study, involved irony (Wang et al., 2007). However, the other study, in using a reading paradigm that required the attribution of complex which also scenarios that involved irony were presented, used a mental states contrasted with rest states, demonstrated more right paradigm that demanded explicit attention to socially relevant frontal activation in the autism group, also indicating a reversal of clues, and found greater activation of the prefrontal and temporal hemisphere dominance (Takeuchi et al., 2004). These studies regions (Wang et al., 2006). should, however, be considered exploratory because of their small Thirdly, the task indifferent pattern of activation during sample sizes and multiple comparisons. language tasks in autism supports the idea that the neural In an fMRI study by Harris et al., the semantic and perceptual networks that are temporarily recruited for a cognitive task cannot processing of single words was assessed (Harris et al., 2006). be reset as easily as in controls when the task is changed. There is Semantic processing (evaluate words as positive or negative) and theoretical evidence that suggests that less flexible network perceptual processing (lower or UPPER case) were contrasted as regrouping could be the result of abnormalities in low-level well as processing of concrete and abstract words. The participants sensory processing (Gustafsson, 1997) in the sense that high-level with autism had weaker left and right frontal activation, but processes are flooded with irrelevant information from low-level greater temporal activation for semantic processing, relative to centres and the extra processing demand impedes the flexibility perceptual task conditions. More importantly, compared with the required for neural assemblies to form (Belmonte et al., 2004b). As control group, there was little differential activation between such, less flexible network regrouping has been associated with semantic and perceptual processing in the autism group. The models of aberrant connectivity in autism. This matter will be concrete versus abstract contrast also demonstrated less activation further discussed below. Interestingly, less flexible network differences between the two conditions in the autism group. regrouping might also be involved in the part of the triad of In another fMRI study, abnormal Broca’s and Wernicke’s area symptoms that involves rigidness and repetitive/restricted beha- (Wernicke’s area is situated in the posterior section of the superior viour. The correspondence between the triad of symptoms in temporal gyrus and the Sylvian fissure) activation in autism was relation to their putative causes will be also addressed below. also demonstrated using a sentence comprehension task with Fourthly, autism is a heterogeneous disorder and the same can syntactically demanding probes (Just et al., 2004). The autism be said about the neuroimaging results on language in autism. group had less activation of Broca’s area and adjacent areas and Some of the conflicting findings can be explained by the wide range more Wernicke’s area activation. Furthermore, the study showed a of phenotypic variation with different intellectual levels and age reduced synchronisation of neural activation across the large-scale groups studied. As a consequence of the difficulty to obtain large cortical network for language processing. The greater Wernicke’s enough sample sizes, only a minority of the neuroimaging studies area activation was interpreted as a tendency for more extensive have evaluated the results at a search volume correct p-value and processing of the meanings of the individual words that make up a have used a random effects analysis and a direct group comparison. sentence in autism. The authors hypothesised that the reduced Broca’s area activation may reflect a reduced ability to integrate 5. Discussion the meaning of individual words into a coherent conceptual and syntactic structure. This may still hold, but Harris’s study showed Over two decades of research have shown that linguistic deficits that Broca’s area and Wernicke’s area are also abnormally activated are present in the majority of individuals with autism. Neurophy- during the processing of single words when no syntactic demands siological studies that sought the neural basis of these deficits have are made (Harris et al., 2006). shown abnormalities at a basic, early level of sensory information The findings on underconnectivity were elaborated in an fMRI processing in autism (Boddaert et al., 2003; Ceponiene et al., 2003). study in which autistic participants processed sentences with a Quite likely, these findings reflect developmental abnormalities, high-imagery or low-imagery content (Kana et al., 2006). Measures and as such, they are in accordance with hypotheses on impaired of functional connectivity showed a reduced functional synchro- integration of cortical information in autism. Basically, these nisation between the frontal and parietal areas. Again, the controls theories state that cortical regions do not operate in synchrony, but had greater Broca’s area and adjacent area activation. In contrast to show disorganised and inadequately selective development of the control group, there were little activation differences between connectivity (Rippon et al., 2007; Just et al., 2004; Belmonte et al., the two conditions in the autism group: the autism group seemed 2004a). Especially impaired connectivity between the frontal lobe to process high- and low-imagery sentences similarly. and other systems has been implicated in autism (Courchesne 1422 W.B. Groen et al. / Neuroscience and Biobehavioral Reviews 32 (2008) 1416–1425 et al., 2005). A growing body of evidence lends support for aberrant constellation associated with communicative abilities contribute connectivity in autism, including deficits in physical connectivity, to the other behavioural domains in autism (social interaction and such as histopathology findings of neuroinflammatory microglial restricted and repetitive behaviours and interests) has been the activation (Vargas et al., 2005), the abnormal developmental topic of an ongoing debate. This issue will be further discussed trajectory of brain size (Redcay and Courchesne, 2005), decreases below. in white matter structure integrity measured with DTI (Keller et al., 2007; Barnea-Goraly et al., 2004; Alexander et al., 2007) and 5.1. The relationship between linguistic impairments and social and deficits in computative connectivity measured as the synchrony of restricted/repetitive impairments in autism time series of activation of cortical regions (Kana et al., 2006; Wilson et al., 2006). Although it is tempting to ascribe a The core deficits in autism are traditionally viewed as unidirectional causal relation to deficits in connectivity and inherently intertwined, i.e. communicative, social and restricted/ impairments in autism such as abnormal language development, repetitive difficulties influence each other and are highly inter- a bi-directional relation between connectivity and functional dependent. Indeed, the acquisition of language is one of the impairments in autism is more likely. In healthy adults, white strongest predictors of long-term positive academic and social matter integrity changes over time (Snook et al., 2005), suggesting outcome in children with autism (Gillberg, 1991). Conversely, joint that experiences influence neural connectivity. The socio-com- attention and immediate imitation predict language ability at ages municative deficits in autism could likewise give rise to abnormal 3–4, and toy play and deferred imitation predict communication connectivity. development at ages 4–6 (Toth et al., 2006). However, recent Yet, it has been argued that, in autism, genetic factors in behavioural-genetic data from a large general population-based interaction with environmental influences lead to an abnormal twin-pair study indicate that phenotypic correlations between neural architecture that causes either increased or reduced neural traits usually associated with autism (social interaction, repetitive- connectivity, or both (Belmonte et al., 2004b). Both conditions restrictive behaviours and communication) are low (Ronald et al., could cause an abnormally low signal-to-noise ratio in developing 2006). Even social interaction and communicative abilities were neural assemblies, given the fact that there is excess noise in only modestly clustered with correlations of 0.2–0.4 (Happe et al., hyperconnected systems and the signal gets lost in the noise in 2006). Furthermore, although each aspect of the triad of hypoconnected systems. This may result in a failure to delimit impairments was found to be highly heritable, cross-trait, cross- activation in perceptual processing centres, forcing higher twin correlations were low, suggesting that most genetic effects processing mechanisms to actively suppress irrelevant sensory are specific to one of the core symptoms. information at a later, less efficient stage. These compensatory One might argue that these behavioural-genetic findings mechanisms can presumably not be reset as quickly as the normal suggest that the neural underpinning of the separate parts of mechanisms of selective attention. Empirically, the allocation of the triad of symptoms is trait-specific as well. However, a different neural resources during higher-order language tasks does indeed pattern of genetic and environmental causal factors for each of the seem to be more task-indifferent in autistic individuals than in symptom domains of autism does not preclude the possibility of a controls. The observation of less flexible network regrouping in common neural deficit. In fact, each of these causal patterns may autism during different language task conditions (see also Section lead to a common neural deficit which further compromises other 4.3) can possibly be explained by the high neural demands that neural functions and give rise to the triad of symptoms in various active filtering of relevant stimuli imposes (Belmonte et al., 2004b). degrees. The neural circuitry that makes up the social brain More direct evidence suggesting reduced cortico–cortico consists of a specific network comprising the amygdala, medial connectivity during language tasks in collaborating cortical areas prefrontal cortex, cingulate cortex, right somatosensory-related has been provided as well (Just et al., 2004; Kana et al., 2006). cortex and temporal lobe structures, including the superior Intraregional underconnectivity during a simple auditory task has temporal sulcus and fusiform gyrus (Adolphs, 2001). The language also been reported (Wilson et al., 2006). The high-level processing brain, on the other hand, comprises the superior temporal lobe, deficits of linguistic stimuli might arise when an unfiltered flood of inferior parietal, temporal-parietal-occipital junction and inferior linguistic information reaches higher information processing units frontal areas (Hickok, 2001; Vigneau et al., 2006). Thus, the speech- that have to evaluate and actively suppress irrelevant sensory specific superior temporal areas are the main link between the inputs. The greater reliance on Wernicke’s area for linguistic social and language networks. The triad of symptoms may not only processing in autism (Just et al., 2004) might thus result from share anatomical characteristics, but have neurophysiological compensatory developmental processes that lead to a more self- characteristics such as impaired connectivity in common as well. reliant Wernicke’s area. Further genetic and neurophysiological studies will need to Additional empirical findings from a number of histopathology address the attributions and, if possible, the direction of causality studies of autistic brains that have shown an atypical distribution of impaired connectivity for the separate parts of the triad. of interneurons suggest that lateral inhibition may be involved in the abnormal neural architecture in autism (Casanova et al., 2003; 5.2. Genetic, endophenotypical and behavioural correspondence Casanova, 2006). Lateral inhibition refers to a physiological process between autism and language impairments in which neurons of one neuronal layer make inhibitory synaptic connections through interneurons in minicolumns. Based on In DSM-IV, there is an explicit contrast between autism and computer simulations, it has been predicted that alterations in specific language impairment. Children with SLI have a specific the strength of lateral inhibition can have a positive effect on disability to acquire age-appropriate language, while development perceptual discrimination abilities but have a detrimental effect on in all other domains is normal (American Psychiatric Association, more global perception, exactly as is the case in autism 1994). Children with autism, by contrast, may not only have a (Gustafsson, 1997). This hypothesis is however arguably spec- delay in language development, other developmental domains are ulative and further research into this matter is required. affected as well and there are also linguistic deviations that are not It should be noted that linguistic and communicative impair- normal for any stage of development (Bishop, 2002). Nevertheless, ments comprise only one-third of the triad of impairments in the linguistic deficits observed in specific language impairment autism. Whether or not abnormal connectivity and the genetic resemble the language impairment in autism. Children with SLI W.B. Groen et al. / Neuroscience and Biobehavioral Reviews 32 (2008) 1416–1425 1423 usually have limited vocabularies, produce immature speech References sounds and use basic grammatical structures (Newbury et al., Abell, F., Krams, M., Ashburner, J., Passingham, R., Friston, K., Frackowiak, R., Happe, 2005). The profile of language performance among a subgroup of F., Frith, C., Frith, U., 1999. The neuroanatomy of autism: a voxel-based whole children with autism mirrors the profile of SLI: poorer performance brain analysis of structural scans. Neuroreport 10, 1647–1651. on tests of grammatical ability than vocabulary, and difficulties Adolphs, R., 2001. The neurobiology of social cognition. Curr. Opin. Neurobiol. 11, 231–239. with non-word repetition (Tomblin and Zhang, 1999; Dollaghan Alexander, A.L., Lee, J.E., Lazar, M., Boudos, R., Dubray, M.B., Oakes, T.R., Miller, and Campbell, 1998; Kjelgaard and Tager-Flusberg, 2001). J.N.,Lu,J.,Jeong,E.K.,McMahon,W.M.,Bigler,E.D.,Lainhart,J.E.,2007. The overlap in both disorders suggests that autism and SLI Diffusion tensor imaging of the corpus callosum in Autism. Neuroimage 34, 61–73. involve a shared neural substrate and one or more shared genes Allen, G., Muller, R.A., Courchesne, E., 2004. Cerebellar function in autism: func- (Kjelgaard and Tager-Flusberg, 2001). Regarding the possibility of a tional magnetic resonance image activation during a simple motor task. Biol. shared genetic aetiology, family members of probands with autism Psychiatry 56, 269–278. have a higher than chance rate of language impairments, and American Psychiatric Association, 1994. Diagnostic and Statistical Manual of Men- tal Disorders. American Psychiatric Association, Washington, DC. family members of probands with SLI are at risk for autism (Piven Asperger, H., 1991. Autistic psychopathy (translation of the original paper by U. and Palmer, 1997; Tomblin et al., 2003). Several genome-wide Frith). In: Frith, U. (Ed.), Autism and Aspger Syndrome. Cambridge University screens have been undertaken to find susceptibility loci for autism, Press, Cambridge. Barnea-Goraly, N., Kwon, H., Menon, V., Eliez, S., Lotspeich, L., Reiss, A.L., 2004. providing the most significant findings on 7q31 (International White matter structure in autism: preliminary evidence from diffusion tensor Molecular Genetic Study of Autism Consortium, 1998) (but for a imaging. Biol. Psychiatry 55, 323–326. review see Klauck, 2006). Baron-Cohen, S., Leslie, A.M., Frith, U., 1985. Does the autistic child have a ‘‘theory of mind’’? Cognition 21, 37–46. Given the similar phenotype and genetic overlap between Baron-Cohen, S., 1997. Hey! It was just a joke! Understanding propositions and autism and SLI, the issue of to what extent underconnectivity is propositional attitudes by normally developing children and children with involved in the emergence of SLI needs to be addressed. autism Isr. J. Psychiatry Relat. Sci. 178. Bartak, L., Rutter, M., Cox, A., 1975. A comparative study of infantile autism and Interestingly, phenotypic similarity at a neural level has been specific development receptive language disorder. I. The children. Br. J. Psy- found between language impaired boys with autism and boys with chiatry 126, 127–145. SLI (DeFosse et al., 2004). Segmented MRI scans showed larger Bartolucci, G., Albers, R.J., 1974. Deictic categories in the language of autistic children. J. Autism Child Schizophr. 4, 131–141. right than left frontal language association cortexes in both groups. Bartolucci, G., Pierce, S., Streiner, D., Eppel, P.T., 1976. Phonological investigation of Since the inversed asymmetry was not present in the autism group verbal autistic and mentally retarded subjects. J. Autism Child Schizophr. 6, that was linguistically unimpaired, these results suggest that there 303–316. is a common underlying (genetic) cause for the language Bartolucci, G., Pierce, S.J., Streiner, D., 1980. Cross-sectional studies of grammatical morphemes in autistic and mentally retarded children. J. Autism Dev. Disord. impairments in autism and SLI. 10, 39–50. Belmonte, M.K., Allen, G., Beckel-Mitchener, A., Boulanger, L.M., Carper, R.A., Webb, 5.3. Implications for therapies and directions for future research S.J., 2004a. Autism and abnormal development of brain connectivity. J. Neurosci. 24, 9228–9231. Belmonte, M.K., Cook Jr., E.H., Anderson, G.M., Rubenstein, J.L., Greenough, W.T., There is a pressing need for further research into functional and Beckel-Mitchener, A., Courchesne, E., Boulanger, L.M., Powell, S.B., Levitt, P.R., structural connectivity in autism. The coarse current knowledge Perry, E.K., Jiang, Y.H., DeLorey, T.M., Tierney, E., 2004b. Autism as a disorder of neural information processing: directions for research and targets for therapy. on neural connectivity could be refined by combining measures of Mol. Psychiatry 9, 646–663. connectivity such as DTI, MEG and fMRI and genotyping (imaging Bertone, A., Mottron, L., Jelenic, P., Faubert, J., 2005. Enhanced and diminished visuo- genomics), so that genes associated with connectivity can be spatial information processing in autism depends on stimulus complexity. Brain 128, 2430–2441. found. Also, if reliable and valid subgroups can be formed based on Bishop, D.V., 2002. Speech and language difficulties. In: Rutter, M., Taylor, E. neuroimaging measures, the power of whole genome screens (Eds.), Child and Adolescent Psychiatry. Blackwell, Malden. would increase considerably. Imaging should also be combined Boddaert, N., Belin, P., Chabane, N., Poline, J.B., Barthelemy, C., Mouren-Simeoni, M.C., Brunelle, F., Samson, Y., Zilbovicius, M., 2003. Perception of complex with administration of pharmacologic agents to allow for sounds: abnormal pattern of cortical activation in autism. Am. J. Psychiatry inferences on the biochemical basis of autism. Apart from 160, 2057–2060. combining investigative modalities, research on autism will also Boddaert, N., Chabane, N., Belin, P., Bourgeois, M., Royer, V., Barthelemy, C., Mouren- benefit from using well-selected subject groups. Knowledge on Simeoni, M.C., Philippe, A., Brunelle, F., Samson, Y., Zilbovicius, M., 2004a. Perception of complex sounds in autism: abnormal auditory cortical processing the developmental trajectory of connectivity in autism could be in children. Am. J. Psychiatry 161, 2117–2120. gained by using different age groups (cross sectional studies) or Boddaert, N., Chabane, N., Gervais, H., Good, C.D., Bourgeois, M., Plumet, M.H., better, by employing longitudinal imaging study designs. Barthelemy, C., Mouren, M.C., Artiges, E., Samson, Y., Brunelle, F., Frackowiak, R.S., Zilbovicius, M., 2004b. Superior temporal sulcus anatomical abnormalities Comparison between people with HFA, Asperger syndrome and in childhood autism: a voxel-based morphometry MRI study. Neuroimage 23, SLI will allow for inferences on language deficits compared to 364–369. social deficits, and the influence of symptom severity on micro Bookheimer, S., 2002. Functional MRI of language: new approaches to understanding the cortical organization of semantic processing. Annu. Rev. Neurosci. 25, architectural measures could come from contrasting people with 151–188. autism and PDD-NOS. Boucher, J., 2003. Language development in autism. Int. J. Pediatr. Otorhinolaryngol. There is furthermore a need for biological markers of autism, 67 (Suppl. 1), S159–S163. Bruneau, N., Roux, S., Adrien, J.L., Barthelemy, C., 1999. Auditory associative cortex since early diagnosis based solely on observation of behavioural dysfunction in children with autism: evidence from late auditory evoked symptoms has been proven difficult and inefficient (Groen et al., potentials (N1 wave-T complex). Clin. Neurophysiol. 110, 1927–1934. 2007). Other psychiatric disorders with an onset in early child- Buitelaar, J.K., van der Gaag, R.J., 1998. Diagnostic rules for children with PDD-NOS and multiple complex developmental disorder. J. Child Psychol. Psychiatry 39, hood, such as attention deficit hyperactivity disorder (Murias et al., 911–919. 2007) have also been associated with aberrant connectivity. Buitelaar, J.K., van der Gaag, R.J., Klin, A., Volkmar, F., 1999. Exploring the boundaries Therefore, the specificity and extent of abnormal connectivity in of pervasive developmental disorder not otherwise specified: analyses of data autism should be addressed, so that the possibility of a from the DSM-IV Autistic Disorder Field Trial. J. Autism Dev. Disord. 29, 33–43. Cantwell, D., Baker, L., Rutter, M., 1978. A comparative study of infantile autism and connectivity-based biomarker can be explored. Diffusion weighted specific developmental receptive language disorder-IV. Analysis of syntax and imaging might facilitate early diagnosis if a sufficiently sensitive language function. J. Child Psychol. Psychiatry 19, 351–362. and specific acquisition procedure can be developed. Early Carper, R.A., Moses, P., Tigue, Z.D., Courchesne, E., 2002. Cerebral lobes in autism: early hyperplasia and abnormal age effects. Neuroimage 16, 1038–1051. diagnosis may be one way of opening the door for effective Casanova, M.F., 2006. Neuropathological and genetic findings in autism: the sig- therapies. nificance of a putative minicolumnopathy. Neuroscientist 12, 435–441. 1424 W.B. Groen et al. / Neuroscience and Biobehavioral Reviews 32 (2008) 1416–1425

Casanova, M.F., Buxhoeveden, D., Gomez, J., 2003. Disruption in the inhibitory Harris, G.J., Chabris, C.F., Clark, J., Urban, T., Aharon, I., Steele, S., McGrath, L., architecture of the cell minicolumn: implications for autisim. Neuroscientist 9, Condouris, K., Tager-Flusberg, H., 2006. Brain activation during semantic pro- 496–507. cessing in autism spectrum disorders via functional magnetic resonance ima- Ceponiene, R., Lepisto, T., Shestakova, A., Vanhala, R., Alku, P., Naatanen, R., Yaguchi, ging. Brain Cogn. 61, 54–68. K., 2003. Speech-sound-selective auditory impairment in children with autism: Herbert, M.R., Harris, G.J., Adrien, K.T., Ziegler, D.A., Makris, N., Kennedy, D.N., Lange, they can perceive but do not attend. Proc. Natl. Acad. Sci. U.S.A. 100, 5567–5572. N.T., Chabris, C.F., Bakardjiev, A., Hodgson, J., Takeoka, M., Tager-Flusberg, H., Chiron, C., Leboyer, M., Leon, F., Jambaque, I., Nuttin, C., Syrota, A., 1995. SPECT of the Caviness Jr, V.S, 2002. Abnormal asymmetry in language association cortex in brain in childhood autism: evidence for a lack of normal hemispheric asym- autism. Ann. Neurol. 52, 588–596. metry. Dev. Med. Child Neurol. 37, 849–860. Hickok, G., 2001. Functional anatomy of speech perception and speech production: Colle, L., Baron-Cohen, S., Hill, J., 2007. Do children with autism have a theory of psycholinguistic implications. J. Psycholinguist. Res. 30, 225–235. mind? A non-verbal test of autism vs. specific language impairment. J. Autism Hobson, R.P., Lee, A., 1989. Emotion-related and abstract concepts in autistic Dev. Disord. 37, 716–723. people: evidence from the British Picture Vocabulary Scale. J. Autism Dev. Courchesne, E., Karns, C.M., Davis, H.R., Ziccardi, R., Carper, R.A., Tigue, Z.D., Chisum, Disord. 19, 601–623. H.J., Moses, P., Pierce, K., Lord, C., Lincoln, A.J., Pizzo, S., Schreibman, L., Haas, Howlin, P., 2003. Outcome in high-functioning adults with autism with and without R.H., Akshoomoff, N.A., Courchesne, R.Y., 2001. Unusual brain growth patterns early language delays: implications for the differentiation between autism and in early life in patients with autistic disorder: an MRI study. Neurology 57, 245– Asperger syndrome. J. Autism Dev. Disord. 33, 3–13. 254. International Molecular Genetic Study of Autism Consortium, 1998. A full genome Courchesne, E., Pierce, K., 2005. Why the frontal cortex in autism might be talking screen for autism with evidence for linkage to a region on chromosome 7q. only to itself: local over-connectivity but long-distance disconnection. Curr. Hum. Mol. Genet. 7, 571–578. Opin. Neurobiol. 15, 225–230. Jolliffe, T., Baron-Cohen, S., 2000. Linguistic processing in high-functioning adults Courchesne, E., Redcay, E., Morgan, J.T., Kennedy, D.P., 2005. Autism at the begin- with autism or Asperger’s syndrome. Is global coherence impaired? Psychol. ning: microstructural and growth abnormalities underlying the cognitive and Med. 30, 1169–1187. behavioral phenotype of autism. Dev. Psychopathol. 17, 577–597. Just, M.A., Cherkassky, V.L., Keller, T.A., Minshew, N.J., 2004. Cortical activation and Cox, A., Rutter, M., Newman, S., Bartak, L., 1975. A comparative study of infantile synchronization during sentence comprehension in high-functioning autism: autism and specific developmental receptive language disorder. II. Parental evidence of underconnectivity. Brain 127, 1811–1821. characteristics. Br. J. Psychiatry 126, 146–159. Kana, R.K., Keller, T.A., Cherkassky, V.L., Minshew, N.J., Just, M.A., 2006. Sentence Cromer, R.F., 1981. Developmental language disorders: cognitive processes, seman- comprehension in autism: thinking in pictures with decreased functional tics, pragmatics, phonology, and syntax. J. Autism Dev. Disord. 11, 57–74. connectivity. Brain. Dawson, G., Finley, C., Phillips, S., Lewy, A., 1989. A comparison of hemispheric Kanner, L., 1943. Autistic disturbances of affective contact. Nervous Child. asymmetries in speech-related brain potentials of autistic and dysphasic chil- Kasai, K., Hashimoto, O., Kawakubo, Y., Yumoto, M., Kamio, S., Itoh, K., Koshida, I., dren. Brain Lang. 37, 26–41. Iwanami, A., Nakagome, K., Fukuda, M., Yamasue, H., Yamada, H., Abe, O., Aoki, Dawson, G., Meltzoff, A.N., Osterling, J., Rinaldi, J., Brown, E., 1998. Children with S., Kato, N., 2005. Delayed automatic detection of change in speech sounds in autism fail to orient to naturally occurring social stimuli. J. Autism Dev. Disord. adults with autism: a magnetoencephalographic study. Clin. Neurophysiol. 116, 28, 479–485. 1655–1664. De Villiers, J., De Villiers, P., 1995. Steps in the mastery of sentence complements. In: Keller, T.A., Kana, R.K., Just, M.A., 2007. A developmental study of the structural Paper Presented at the Biennial Meeting of the Society for Research in Child integrity of white matter in autism. Neuroreport 18, 23–27. Development. Kjelgaard, M.M., Tager-Flusberg, H., 2001. An investigation of language impairment DeFosse, L., Hodge, S.M., Makris, N., Kennedy, D.N., Caviness Jr, V.S., McGrath, L., in autism: implications for genetic subgroups. Lang. Cogn. Process 16, 287–308. Steele, S., Ziegler, D.A., Herbert, M.R., Frazier, J.A., Tager-Flusberg, H., Harris, G.J., Klauck, S.M., 2006. Genetics of autism spectrum disorder. Eur. J. Hum. Genet. 14, 2004. Language-association cortex asymmetry in autism and specific language 714–720. impairment. Ann. Neurol. 56, 757–766. Klin, A., 1991. Young autistic children’s listening preferences in regard to speech: a Dollaghan, C., Campbell, T.F., 1998. Nonword repetition and child language impair- possible characterization of the symptom of social withdrawal. J. Autism Dev. ment. J. Speech Lang. Hear. Res. 41, 1136–1146. Disord. 21, 29–42. Dunn, L., Dunn, L., Whetton, C., Burley, J., 1997. British Picture Vocabulary Scale: Klin, A., Pauls, D., Schultz, R., Volkmar, F., 2005. Three diagnostic approaches to Revised. NFER Nelson Publishing Co., Slough, Bucks. Asperger syndrome: implications for research. J. Autism Dev. Disord. 35, 221– Dunn, L.M., Dunn, M., 1997. Peabody Picture Vocabulary Test. American Guidance 234. Service, Circel Pines, MN. Kuhl, P.K., Tsao, F.M., Liu, H.M., 2003. Foreign-language experience in infancy: Eigsti, I.M., Bennetto, L., Dadlani, M.B., 2007. Beyond pragmatics: morphosyntactic effects of short-term exposure and social interaction on phonetic learning. Proc. development in autism. J. Autism Dev. Disord. 37, 1007–1023. Natl. Acad. Sci. U.S.A. 100, 9096–9101. Eskes, G.A., Bryson, S.E., McCormick, T.A., 1990. Comprehension of concrete and Levitt, J.G., Blanton, R.E., Smalley, S., Thompson, P.M., Guthrie, D., McCracken, J.T., abstract words in autistic children. J. Autism Dev. Disord. 20, 61–73. Sadoun, T., Heinichen, L., Toga, A.W., 2003. Cortical sulcal maps in autism. Cereb. Fine, J., Bartolucci, G., Szatmari, P., Ginsberg, G., 1994. Cohesive discourse in Cortex 13, 728–735. pervasive developmental disorders. J. Autism Dev. Disord. 24, 315–329. Lord, C., Paul, R., 1997. Language and communication in autism. In: Cohen, D., Flagg, E.J., Cardy, J.E., Roberts, W., Roberts, T.P., 2005. Language lateralization Volkmar, F.R. (Eds.), Handbook of Autism and Pervasive Developmental Dis- development in children with autism: insights from the late field magnetoen- orders. John Wiley, New York. cephalogram. Neurosci. Lett. 386, 82–87. Martin, I., McDonald, S., 2004. An exploration of causes of non-literal language Frith, U., 1996. Cognitive explanations of autism. Acta Paediatr. Suppl. 416, 63–68. problems in individuals with Asperger Syndrome. J. Autism Dev. Disord. 34, Frith, U., Snowling, M., 1983. Reading for meaning and reading for sound in autistic 311–328. and dyslexic children. Br. J. Dev. Psychol.. Muller, R.A., Behen, M.E., Rothermel, R.D., Chugani, D.C., Muzik, O., Mangner, T.J., Gardner, M., 1982. Expressive One Word Vocabulary Test. Western Psychological Chugani, H.T., 1999. Brain mapping of language and auditory perception in Services, Los Angeles. high-functioning autistic adults: A PET study. J. Autism Dev. Disord. 29, 19–31. Gendry-Meresse, I., Zilbovicius, M., Boddaert, N., Robel, L., Philippe, A., Sfaello, I., Muller, R.A., Chugani, D.C., Behen, M.E., Rothermel, R.D., Muzik, O., Chakraborty, P.K., Laurier, L., Brunelle, F., Samson, Y., Mouren, M.C., Chabane, N., 2005. Autism Chugani, H.T., 1998. Impairment of dentato-thalamo-cortical pathway in autis- severity and temporal lobe functional abnormalities. Ann. Neurol. 58, 466–469. tic men: language activation data from positron emission tomography. Neu- Gervais, H., Belin, P., Boddaert, N., Leboyer, M., Coez, A., Sfaello, I., Barthelemy, C., rosci. Lett. 245, 1–4. Brunelle, F., Samson, Y., Zilbovicius, M., 2004. Abnormal cortical voice proces- Murias, M., Swanson, J.M., Srinivasan, R., 2007. Functional connectivity of frontal sing in autism. Nat. Neurosci. 7, 801–802. cortex in healthy and ADHD children reflected in EEG coherence. Cereb. Cortex Gillberg, C., 1991. Outcome in autism and autistic-like conditions. J. Am. Acad. Child 17, 1788–1799. Adolesc. Psychiatry 30, 375–382. Newbury, D.F., Bishop, D.V., Monaco, A.P., 2005. Genetic influences on language Groen, W.B., Swinkels, S.H., van der Gaag, R.J., Buitelaar, J.K., 2007. Finding effective impairment and phonological short-term memory. Trends Cogn. Sci. 9, 528– screening instruments for autism using bayes theorem. Arch. Pediatr. Adolesc. 534. Med. 161, 415–416. Oram Cardy, J.E., Flagg, E.J., Roberts, W., Roberts, T.P., 2005. Delayed mismatch field Gustafsson, L., 1997. Inadequate cortical feature maps: a neural circuit theory of for speech and non-speech sounds in children with autism. Neuroreport 16, autism. Biol. Psychiatry 42, 1138–1147. 521–525. Hagoort, P., Hald, L., Bastiaansen, M., Petersson, K.M., 2004. Integration of word Palmen, S.J., van Engeland, H., 2004. Review on structural neuroimaging findings in meaning and world knowledge in language comprehension. Science 304, 438– autism. J. Neural. Transm. 111, 903–929. 441. Paul, R., Augustyn, A., Klin, A., Volkmar, F.R., 2005. Perception and production of Happe, F., Frith, U., 2006. The weak coherence account: detail-focused cognitive prosody by speakers with autism spectrum disorders. J. Autism Dev. Disord. 35, style in autism spectrum disorders. J. Autism Dev. Disord. 36, 5–25. 205–220. Happe, F., Ronald, A., Plomin, R., 2006. Time to give up on a single explanation for Perkins, M.R., Dobbinson, S., Boucher, J., Bol, S., Bloom, P., 2006. Lexical knowledge autism. Nat. Neurosci. 9, 1218–1220. and lexical use in autism. J. Autism Dev. Disord. 36, 795–805. Happe, F.G., 1993. Communicative competence and theory of mind in autism: a test Piven, J., Palmer, P., 1997. Cognitive deficits in parents from multiple-incidence of relevance theory. Cognition 48, 101–119. autism families. J. Child Psychol. Psychiatry 38, 1011–1021. Happe, F.G.E., 1997. Central coherence and theory of mind in autism: reading Rapin, I., Dunn, M., 2003. Update on the language disorders of individuals on the homographs in context. Br. J. Dev. Psychol. 15, 1–12. autistic spectrum. Brain Dev. 25, 166–172. W.B. Groen et al. / Neuroscience and Biobehavioral Reviews 32 (2008) 1416–1425 1425

Redcay, E., Courchesne, E., 2005. When is the brain enlarged in autism? A meta- Takeuchi, M., Harada, M., Matsuzaki, K., Nishitani, H., Mori, K., 2004. Difference of analysis of all brain size reports. Biol. Psychiatry 58, 1–9. signal change by a language task on autistic patients using functional MRI. J. Rippon, G., Brock, J., Brown, C., Boucher, J., 2007. Disordered connectivity in the Med. Invest. 51, 59–62. autistic brain: challenges for the ‘new psychophysiology’. Int. J. Psychophysiol. Tantam, D., Holmes, D., Cordess, C., 1993. Nonverbal expression in autism of 164–172. Asperger type. J. Autism Dev. Disord. 23, 111–133. Roberts, J.A., Rice, M.L., Tager-Flusberg, H., 2004. Tense marking in children with Tomblin, J.B., Hafeman, L.L., O’Brien, M., 2003. Autism and autism risk in siblings of autism. Appl. Psycholinguist. 25, 429–448. children with specific language impairment. Int. J. Lang. Commun. Disord. 38, Ronald, A., Happe, F., Price, T.S., Baron-Cohen, S., Plomin, R., 2006. Phenotypic 235–250. and genetic overlap between autistic traits at the extremes of the Tomblin, J.B., Zhang, X., 1999. Language patterns and etiology in children with general population. J. Am. Acad. Child Adolesc. Psychiatry 45, 1206– specific language impairment. In: Tager-Flusberg, H. (Ed.), Neurodevelop- 1214. mental Disorders. MIT Press/Bradford Books, Cambridge, MA. Rutherford, M.D., Baron Cohen, S., Wheelwright, S., 2002. Reading the mind in the Toth, K., Munson, J., Meltzoff, A.N., Dawson, G., 2006. Early predictors of commu- voice: a study with normal adults and adults with Asperger syndrome and high nication development in young children with autism spectrum disorder: joint functioning autism. J. Autism Dev. Disord. 32, 189–194. attention, imitation, and toy play. J. Autism Dev. Disord. 36, 993–1005. Shriberg, L.D., Paul, R., McSweeny, J.L., Klin, A.M., Cohen, D.J., Volkmar, F.R., 2001. Vargas, D.L., Nascimbene, C., Krishnan, C., Zimmerman, A.W., Pardo, C.A., 2005. Speech and prosody characteristics of adolescents and adults with high-func- Neuroglial activation and neuroinflammation in the brain of patients with tioning autism and Asperger syndrome. J. Speech Lang. Hear. Res. 44, 1097– autism. Ann. Neurol. 57, 67–81. 1115. Venter, A., Lord, C., Schopler, E., 1992. A follow-up study of high-functioning autistic Smitz, C., 2005. New aspects in the neuropathology of the forebrain in autism. In: children. J. Child Psychol. Psychiatry 33, 489–507. Paper Presented at the Integrating the Clinical and Basic Sciences of Autism: A Vigneau, M., Beaucousin, V., Herve, P.Y., Duffau, H., Crivello, F., Houde, O., Mazoyer, Developmental Biology Workshop. B., Tzourio-Mazoyer, N., 2006. Meta-analyzing left hemisphere language areas: Snook, L., Paulson, L.A., Roy, D., Phillips, L., Beaulieu, C., 2005. Diffusion tensor phonology, semantics, and sentence processing. Neuroimage 30, 1414–1432. imaging of neurodevelopment in children and young adults. Neuroimage 26, Volkmar, F.R., Lord, C., Bailey, A., Schultz, R.T., Klin, A., 2004. Autism and pervasive 1164–1173. developmental disorders. J. Child Psychol. Psychiatry 45, 135–170. Tager Flusberg, H., 1981. On the nature of linguistic functioning in early infantile Wang, A.T., Lee, S.S., Sigman, M., Dapretto, M., 2006. Neural basis of irony com- autism. J. Autism Dev. Disord. 11, 45–56. prehension in children with autism: the role of prosody and context. Brain 129, Tager Flusberg, H., 1996. Brief report: current theory and research on language and 932–943. communication in autism. J. Autism Dev. Disord. 26, 169–172. Wang, A.T., Lee, S.S., Sigman, M., Dapretto, M., 2007. Reading affect in the face and Tager Flusberg, H., Sullivan, K., 1995. Attributing mental states to story characters: a voice: neural correlates of interpreting communicative intent in children and comparison of narratives produced by autistic and mentally retarded indivi- adolescents with autism spectrum disorders. Arch. Gen. Psychiatry 64, 698–708. duals. Appl. Psycholinguist.. Williams, K.T., 1997. Expressive Vocabulary Test. American Guidance Service, Circle Tager-Flusberg, H., 2000. Language and understanding minds: connections in Pines, MN. autism. In: Baron-Cohen, S., Tager-Flusberg, H., Cohen, D. (Eds.), Understand- Wilson, T.W., Rojas, D.C., Reite, M.L., Teale, P.D., Rogers, S.J., 2006. Children and ing Other Minds: Perspectives From Developmental Cognitive Neuroscience. adolescents with autism exhibit reduced MEG steady-state gamma responses. Oxford. Biol. Psychiatry.